Empirical study measuring hallucination, confidence, and bias in large language models under prompt stress.
SADI et al. (Thu,) studied this question.